Albumin, Dose-Response, and Anastomotic Leakage in GI Surgery - A Meta-Analysis Framework

1. Biological Rationale - Why Albumin Matters for Anastomotic Healing

2. Existing Evidence Landscape

What already exists

• Observational cohort studies (retrospective or prospective) reporting hypoalbuminemia as a univariable or multivariable risk factor for AL
• Composite risk factor meta-analyses (e.g., McDermott et al., 2015, Br J Surg, PMID 25703524) that include albumin as one of many predictors
• Nutritional intervention meta-analyses (ESPEN 2025: 29 RCTs, 2552 patients after GI surgery showing EN reduces anastomotic insufficiency) - but these use albumin as a mediating outcome, not the exposure

What is critically missing - your niche

1. Positions serum albumin level specifically as the primary exposure variable for AL
2. Explores the dose-response relationship (e.g., each 1 g/dL drop in albumin → incremental change in AL risk)
3. Differentiates preoperative vs. postoperative albumin trajectories as distinct exposures
4. Stratifies across GI surgery subtypes (esophageal, gastric, hepatopancreatic, colorectal, small bowel)

3. Feasibility Assessment

Arguments FOR feasibility

Arguments AGAINST / Caution flags

4. Eligibility Criteria - Explicit Framework

Population (P)

• Include: Adult patients (≥18 years) undergoing elective or emergency GI surgery with formation of a GI anastomosis (esophagogastric, gastrojejunal, ileocolic, colorectal, colo-anal, hepaticojejunal, pancreaticojejunal)
• Exclude: Pediatric patients; non-anastomotic surgeries (stoma formation only); organ transplantation; vascular anastomoses; esophageal reconstruction with non-GI conduits (e.g., cutaneous tubes)
• Subgroup pre-specification: colorectal vs. upper GI vs. hepatopancreaticobiliary - because AL rates, biology, and albumin dynamics differ substantially

Exposure/Intervention (I)

• Primary: Serum albumin level measured preoperatively (within 30 days of surgery), defined as continuous variable (g/dL or g/L) OR categorical (hypoalbuminemia vs. normal, using author-defined or pre-specified thresholds)
• Secondary exposures to explore:
  • Postoperative day 1-3 albumin (acute drop)
  • Delta albumin (preoperative minus postoperative)
  • Albumin trajectory (serial measurements)
• Dose-response arm: Studies must report albumin as a continuous variable, OR provide sufficient data (mean, SD, number of AL events per albumin stratum) to reconstruct a dose-response curve
• Exclude: Albumin infusion/supplementation as the intervention (different question - that is a therapeutic meta-analysis)

Comparator (C)

• Normal serum albumin (>35 g/L or author-defined)
• Or: higher albumin level (for dose-response arm - no explicit comparator needed)

Outcome (O)

• Primary: Anastomotic leakage/failure (any grade), defined according to:
  • ISREC (International Study Group of Rectal Cancer) criteria
  • Clinical (re-operation, peritonitis, fecal/enteric drain output)
  • Radiological (CT-confirmed leak/abscess)
  • Endoscopic confirmation
• Secondary outcomes (to add relevance to reviewers):
  • Grade C AL (requiring re-operation or resulting in death)
  • 30-day mortality
  • Hospital/ICU length of stay
  • Surgical site infection
  • 90-day readmission
• Outcome ascertainment period: minimum 30 days postoperative

Study Design (S)

• Include: Cohort studies (prospective and retrospective), case-control studies, RCTs (if albumin data reported as exposure alongside a randomized intervention), cross-sectional studies with sufficient follow-up
• Exclude: Case reports, case series (<20 patients), abstracts only, conference proceedings without full data, studies with no outcome data extractable for anastomotic leak specifically
• Language: English + major European languages (with translation if needed) - narrow to English-only only if resources don't allow translation, but acknowledge as limitation
• Date range: No restriction on start date; up to current search date. Sensitivity analysis limiting to post-2000 (modern anastomotic techniques era) is advisable.
• Minimum sample size: Pre-specify n ≥ 50 anastomoses (to avoid extreme small-study effects)

5. Critical Appraisal Tools

For observational studies (majority of your corpus)

• Selection (4 items): representativeness, comparability of exposed/unexposed
• Comparability (1 item): control for confounders - critical for albumin because you must check whether studies adjusted for age, cancer stage, ASA grade, neoadjuvant therapy, diabetes, steroid use, and operative blood loss
• Outcome (3 items): blinded assessment, follow-up adequacy

1. Confounding bias: Did the study adjust for the major confounders of AL (listed above)? Albumin in many studies is reported only in univariable analysis - these should be flagged or excluded from the main analysis
2. Measurement bias: Was albumin measured by the same assay? Bromocresol green vs. bromocresol purple methods differ by ~3-4 g/L - this is a real source of systematic error
3. Outcome ascertainment bias: Was AL adjudicated by a blinded reviewer? Were asymptomatic leaks detected on routine CT or only symptomatic leaks reported?
4. Selection bias: Were patients with pre-existing liver disease (confounds albumin) or renal disease excluded?

For RCTs (if any meet criteria)

Grading overall evidence

6. Heterogeneity - The Central Methodological Challenge

Sources of clinical heterogeneity

Sources of methodological heterogeneity

Management strategies

1. Surgery type: colorectal, upper GI, HPB separately
2. Albumin timing: preoperative vs. postoperative
3. AL definition: clinical vs. radiological vs. ISREC
4. Adjusted vs. unadjusted effect estimates
5. Emergency vs. elective surgery

• Restrict to studies with NOS ≥6 (moderate-high quality)
• Restrict to studies with multivariable-adjusted ORs only
• Restrict to studies with standardised AL definition
• Leave-one-out analysis (to identify influence of dominant studies)
• Restrict to post-2010 (ERAS era)

• Use random-effects model (DerSimonian-Laird or REML) as default - do NOT use fixed effects given expected high heterogeneity
• Report I², τ², and prediction intervals (PI). The PI is more informative than I² alone - if PI crosses 1.0, heterogeneity is so large that the pooled estimate may not apply to a given clinical context
• For dose-response: use one-stage dose-response meta-analysis (Greenland-Longnecker method or restricted cubic splines - requires individual study data or stratum-specific data). This is methodologically sophisticated and strengthens the contribution.
• Consider meta-regression with albumin threshold, surgery type, and study year as moderators

7. Reasons for Rejection in Journals - Pre-emption Guide

Reason 1: "The question is already answered"

Reason 2: "Overwhelming heterogeneity makes pooling meaningless"

• Pre-specify that your primary analysis is subgroup-specific (colorectal only, or preoperative albumin only) rather than a grand pool
• Frame the meta-analysis as a quantitative synthesis of heterogeneity sources rather than a single effect estimate
• Use narrative synthesis for high-heterogeneity subgroups

Reason 3: "Albumin is a confounder, not a cause - this is methodologically flawed"

• Explicitly stratify by whether studies adjusted for major confounders
• Include a mechanistic section (Section 1 above) to justify biological plausibility
• Frame albumin as a prognostic biomarker rather than causal risk factor if dose-response is the goal
• Cite the ESPEN 2025 recommendation that albumin defines "high metabolic risk" patients

Reason 4: "Insufficient number of studies for dose-response analysis"

• Conduct a scoping search first (publish a scoping review or protocol) to confirm data availability before committing
• If insufficient for dose-response, pivot to a categorical meta-analysis (hypoalbuminemia yes/no) as the main analysis
• Register a PROSPERO protocol specifying both approaches with pre-defined decision rules

Reason 5: "PROSPERO registration absent or retrospective"

Reason 6: "PRISMA 2020 non-compliance"

Reason 7: "Language and publication bias not adequately addressed"

• Search MEDLINE, Embase, Cochrane, CINAHL, and clinicaltrials.gov
• Include Egger's test and funnel plot asymmetry
• Use trim-and-fill method or Copas selection model if asymmetry detected

Reason 8: "Target journal scope mismatch"

8. Scope and Nature of the Review - Final Position Statement

9. Registration, Reporting, and Publication Strategy Summary

Association Between Preoperative Hypoalbuminemia and Anastomotic Leak After Colorectal Surgery Stratified by Albumin Threshold: A Dose-Response Meta-Analysis

Confirmed Gap in the Literature

• No dedicated meta-analysis exists on preoperative hypoalbuminemia as the primary exposure and AL as the primary outcome in colorectal surgery specifically
• The closest published work is the Dias et al. 2022 meta-analysis (PMID 36449942) on preoperative risk factors for AL after colectomy for colorectal cancer - it identified 14 risk factors but did not include albumin in its pooled analysis (albumin was screened but excluded due to insufficient comparable data)
• The Elfadil et al. 2026 systematic review (PMID 41913878, PROSPERO CRD420251034523) synthesised nutritional status and AL across 32 studies and found hypoalbuminemia "consistently associated with increased risk" - but rated that evidence as moderate certainty only, did not run a dose-response analysis, and did not stratify by albumin threshold
• The Hasil et al. 2026 systematic review (PMID 41401601) examined nutrition interventions vs. AL - concluded current evidence is "insufficient to draw firm conclusions"
• The Venn et al. 2023 systematic review of prediction scores (PMID 37463818) found albumin appears as a predictor in only 1 of 9 existing AL prediction scores (11.1%), despite biological plausibility

1. Title and PICOS Frame

2. Explicit PICOS Eligibility Criteria

Population (P) - Full Specification

• Adults (≥18 years)
• Undergoing elective or emergency colorectal resection surgery
• With creation of a primary colorectal anastomosis (any type: end-to-end, end-to-side, side-to-side, stapled or hand-sewn)
• Surgery types: right hemicolectomy, left hemicolectomy, sigmoid colectomy, anterior resection (high, low, ultra-low), abdominoperineal resection with pull-through, subtotal/total colectomy, Hartmann reversal with anastomosis
• Indication: colorectal cancer, benign disease (diverticular disease, inflammatory bowel disease, volvulus), or mixed populations
• Any surgical approach: open, laparoscopic, robotic

• Pediatric patients (<18 years)
• Emergency Hartmann's procedure (no anastomosis formed)
• Intestinal stoma formation without anastomosis
• Non-colorectal GI surgery (esophageal, gastric, small bowel, HPB) - this keeps the review homogeneous and avoids one of the principal rejection reasons
• Liver transplantation or multi-organ surgery
• Ileostomy or colostomy reversal without colorectal anastomosis
• Studies in which albumin was measured only postoperatively (postoperative albumin reflects surgical stress, not nutritional exposure)
• Studies where albumin was measured more than 60 days preoperatively (risk of interim change)

Exposure (I/E) - Full Specification

• Continuous (g/dL or g/L, with means/SDs or medians/IQR in leak vs. no-leak groups), OR
• Categorical, using any pre-specified threshold (common thresholds: <2.0, <2.5, <3.0, <3.5, <3.8, <4.0 g/dL)

• Albumin as a continuous variable with sufficient statistical data (mean ± SD by outcome group, or regression coefficient per unit albumin)
• Two or more categorical strata of albumin with event counts (e.g., <3.0 vs. 3.0-3.5 vs. >3.5 g/dL)
• OR: sufficient raw data to calculate stratum-specific rates for reconstruction of a dose-response curve

Comparator (C)

• Normal serum albumin (above study-defined or pre-specified threshold)
• For dose-response: higher albumin concentration serves as implicit comparator - no discrete comparator required
• Reference category for dose-response: albumin ≥3.5 g/dL (35 g/L) - the most widely used clinical threshold

Outcome (O) - Primary and Secondary

• Clinical: fever, peritonitis, purulent/fecal drainage, localised abscess
• Radiological: CT or contrast enema demonstrating extravasation, anastomotic defect, or peri-anastomotic collection
• Operative: defect identified at re-laparotomy or endoscopy
• Classification-based: ISREC grade A, B, or C (preferred - allows severity stratification)

• Grade C (operative/fatal) AL specifically
• 30-day all-cause mortality
• Surgical site infection (deep or organ/space)
• Hospital length of stay
• Unplanned re-operation rate
• 90-day readmission

Study Design (S)

• Prospective cohort studies
• Retrospective cohort studies (including registry studies)
• Case-control studies (if sufficient data on albumin levels are extractable)
• RCTs reporting albumin as a baseline covariate with outcome data for AL

• Case series (<50 anastomoses - to prevent extreme small-study effects in a rare outcome)
• Cross-sectional studies (cannot establish temporal precedence of albumin before AL)
• Case reports
• Editorials, letters, conference abstracts (no full peer-reviewed data)
• Studies with missing raw or extractable data on both exposure and outcome

3. The Dose-Response Design - Specifics

What "dose-response" means here

Data types required

Methodological approach for dose-response

• Assign median or midpoint albumin value to each stratum
• Assign a reference category (albumin ≥3.5 g/dL)
• Log-transform ORs and variance from each stratum
• Pool across studies accounting for within-study correlation between strata

1. Linear (one parameter - slope)
2. Restricted cubic splines (3 knots at 10th, 50th, 90th percentile of albumin distribution; 2 parameters) - allows detection of non-linearity
3. Quadratic (2 parameters)

Pre-specified threshold strata for stratified analysis

4. Heterogeneity - Sources, Quantification, and Management

Expected heterogeneity: High

Clinical heterogeneity sources - complete taxonomy

Statistical approach to heterogeneity

1. Quantification: Report I² (with 95% CI using Higgins' method), τ² (REML estimator preferred over DL for small meta-analyses), and the 95% prediction interval alongside the pooled estimate. The prediction interval is mandatory - it tells clinicians the range of true effects in a new study setting and is required by PRISMA 2020.
2. Model: Use random-effects model (REML) as the primary estimator. Report fixed-effects as sensitivity analysis. Never use DerSimonian-Laird as primary if <10 studies (underpowers τ² estimation).
3. Subgroup analyses: All pre-registered. Perform test for subgroup differences (χ² between-subgroup test). Do NOT interpret within-subgroup estimates as definitive if the between-subgroup test is non-significant.
4. Meta-regression: Use for continuous moderators (mean study albumin level, year of publication, proportion with neoadjuvant therapy, proportion with rectal surgery). Requires minimum 10 studies per covariate analysed (Borenstein rule).
5. Publication bias: Funnel plot + Egger's test (minimum 10 studies required for valid interpretation). If asymmetry detected, use trim-and-fill (Duval-Tweedie) and Copas selection model to estimate bias-corrected effect. Report both.

5. Critical Appraisal Framework

Tool selection by study type

NOS scoring for this specific question - mandatory confounder adjustment items

GRADE evidence profile

• Large effect size (pooled OR ≥2.0)
• Dose-response gradient demonstrated (p<0.05 for trend)
• All plausible confounders would only attenuate the association

• Serious risk of bias (majority of studies NOS <6)
• Serious inconsistency (I² >75%, PI crosses 1.0)
• Indirectness (e.g., only cancer patients studied, not generalizable to IBD)
• Imprecision (95% CI includes clinical null effect, i.e., crosses OR 1.0 or CI very wide)
• Publication bias (Egger's p<0.05)

6. Feasibility Assessment - Specific to This Title

Positive feasibility signals

• Primary literature volume: The preoperative risk factor literature for colorectal AL is large. The Dias 2022 meta-analysis searched 15 years and found 64 eligible studies on preoperative risk factors. Most of these reported albumin in at least univariable analysis. Realistically, you can expect 30-60 studies reporting preoperative albumin and AL in colorectal surgery.
• Existing infrastructure: The ISREC grading system (Rahbari et al., 2010, Surgery) is widely adopted since 2010, reducing outcome heterogeneity in more recent studies.
• ERAS guideline alignment: ERAS Society 2023 colorectal guidelines identify preoperative albumin as a nutritional risk marker. Your meta-analysis directly informs these guidelines.
• Dose-response data: The Latin American multicentric cohort found albumin <3.5 g/dL with OR 22.2 (95% CI 11.5-42.9) as an independent predictor. Several nomogram-building studies (e.g., Zhang et al. 2022, PMID 36684195) report albumin as a continuous covariate, generating usable regression coefficients.
• No competing PROSPERO protocol: Searches of PROSPERO in June 2026 found no registered protocol specifically on albumin-threshold dose-response for colorectal AL. The Elfadil 2026 SR (PROSPERO CRD420251034523) has already published, confirming no live competition.

Negative feasibility signals and mitigations

7. Reasons for Rejection - Specific to This Exact Title

Rejection reason 1: "This is just a component of the Elfadil 2026 SR"

Rejection reason 2: "Albumin is a confounder, not a cause - the dose-response is biologically spurious"

• Frame albumin explicitly as a prognostic biomarker, not a causal modifiable risk factor. Your title already correctly uses "association"
• Restrict the primary analysis to adjusted ORs only (which control for ASA grade, cancer stage, comorbidities)
• Perform sensitivity analysis limited to elective surgery only (eliminates emergency presentation as a confounding pathway)
• In your Discussion, explicitly acknowledge that albumin-associated AL risk may reflect a causal nutritional pathway (low protein → impaired collagen synthesis), a confounding pathway (cancer/illness → both low albumin and anastomotic failure), or both - and that your design cannot distinguish these

Rejection reason 3: "The dose-response analysis is underpowered or data are insufficient"

• Conduct a scoping search before submitting to PROSPERO to confirm minimum study numbers. You need at least 5 studies with two or more albumin strata for Greenland-Longnecker; ideally 8-10.
• Pre-specify in your protocol: "If fewer than 5 studies provide stratum-specific data for the dose-response analysis, the dose-response arm will be replaced by a narrative synthesis and the primary analysis will be the binary meta-analysis stratified by albumin threshold."
• This contingency plan prevents having to retract the dose-response framing after registration.

Rejection reason 4: "Overwhelming heterogeneity (I² >90%) makes the pooled estimate uninterpretable"

• Report prediction intervals proactively. If they cross 1.0 in the primary analysis, make this a finding, not a failure: "The wide prediction interval (0.x to x.x) reflects genuine variability across surgical settings, consistent with the known context-dependence of albumin-associated AL risk"
• Pivot interpretation to the subgroup analyses (e.g., "In rectal surgery specifically, hypoalbuminemia was consistently associated with AL [OR x.x, 95% PI x.x-x.x] with lower heterogeneity")
• The goal of the review is partly to characterise and explain heterogeneity, not only to produce a single pooled number

Rejection reason 5: "Methodologically inconsistent AL definitions preclude pooling"

• Subgroup analysis by AL definition type (clinical vs. ISREC vs. radiological vs. not specified)
• Sensitivity analysis restricted to ISREC-graded studies only
• The primary finding of outcome definition inconsistency is itself a publishable contribution - include a table documenting AL definition use across included studies (this adds methodological value beyond just the pooled estimates)

Rejection reason 6: "Scope is too narrow for the target journal"

• What albumin threshold should trigger preoperative nutritional optimisation?
• Does the dose-response support a graded approach to stoma formation decisions?
• How should this evidence influence ERAS nutrition protocols?

8. Search Strategy - Verified and Reproducible

Databases

MEDLINE search string (sample - adapt for Embase/CINAHL)

("colorectal surgery"[MeSH] OR "colectomy"[MeSH] OR "rectal surgery"[MeSH] OR 
"anterior resection"[tiab] OR "hemicolectomy"[tiab] OR "sigmoid resection"[tiab] OR 
"colorectal resection"[tiab] OR "rectal resection"[tiab])

AND

("anastomotic leak"[MeSH] OR "anastomotic leakage"[tiab] OR "anastomotic leak"[tiab] 
OR "anastomotic fistula"[tiab] OR "anastomotic failure"[tiab] OR "anastomotic insufficiency"[tiab])

AND

("albumin"[MeSH] OR "serum albumin"[tiab] OR "hypoalbuminemia"[MeSH] OR 
"hypoalbuminaemia"[tiab] OR "hypoalbuminemia"[tiab] OR "low albumin"[tiab] OR 
"preoperative albumin"[tiab] OR "nutritional status"[MeSH])

9. Data Extraction Template

10. Reporting Standards and Target Journal Strategy

Target journals ranked by fit

11. Summary of What Makes This Publishable

1. First dedicated dose-response meta-analysis of albumin and AL in colorectal surgery - this is the primary selling point
2. Threshold stratification (<2.5 vs. <3.0 vs. <3.5 g/dL) produces directly actionable clinical thresholds for preoperative optimisation decisions and stoma formation thresholds
3. Spline curve output (if data support it) allows clinicians to see whether there is a linear, threshold, or saturation effect - which has real implications for how aggressively albumin correction should be pursued preoperatively
4. Methodological contribution: Documenting AL definition heterogeneity and its impact on effect estimates adds standalone value (building on Heuvelings 2024)
5. Guideline alignment: Directly informs ERAS Colorectal and ESPEN Surgery nutrition guidance, both of which cite albumin as a risk marker but lack quantified threshold-specific risk estimates

The Better Meta-Analysis Topic: My Recommendation

The Landscape at a Glance - What Is Already Saturated

My Recommendation

The preoperative C-reactive protein-to-albumin ratio (CAR) as a predictor of anastomotic leakage after colorectal cancer surgery: a systematic review, meta-analysis, and optimal cut-off determination

Why the CAR, Not Albumin Alone

1. It has the biological superiority argument

• Systemic inflammation (CRP, the numerator - rises in cancer, malnutrition, infection)
• Nutritional-hepatic reserve (albumin, the denominator - falls in malnutrition, inflammation, liver compromise)

2. The literature gap is confirmed and clean

• Zero meta-analyses on CAR + anastomotic leak in CRC surgery
• Zero PROSPERO protocols registered for this question
• Multiple single-centre/multicentre cohort studies reporting CAR as a predictor - enough primary literature to power a meta-analysis
• The closest published work (Zhou 2019) covers CAR and CRC survival/prognosis, not anastomotic leak - different outcome entirely

3. Cut-off determination is the dose-response equivalent - and more actionable

4. Methodological positioning is superior

5. Primary data is computable from existing literature

Full Title and Subtitle

What Makes This Different From Everything Published

PICOS Frame (Abbreviated)

The Key Methodological Innovation - Optimal Cut-off Meta-Analysis

• Generate a summary ROC (sROC) curve
• Calculate pooled sensitivity, specificity, LR+, LR-, DOR
• Identify the optimal CAR threshold (Youden index maximisation across studies)

Expected Findings and Their Clinical Impact

Target Journals - Ranked

The One-Sentence Pitch to a Journal Editor

Honesty Check - Potential Obstacles

1. Primary study numbers: If fewer than 8-10 studies report preoperative (not postoperative) CAR with AL outcome, the ROC meta-analysis arm will be underpowered. Do a scoping search first before PROSPERO registration.
2. CRP assay heterogeneity: High-sensitivity CRP (hsCRP) vs. standard CRP report values on different scales. Must be pre-specified as a subgroup variable.
3. Elective CRC only vs. mixed indications: Stick to elective CRC only - this is cleaner and avoids the emergency surgery confounder that killed several albumin meta-analyses.
4. Neoadjuvant therapy: This is actually a strength for your subgroup analysis - you can show whether CAR maintains predictive value even in the post-neoadjuvant patient (where albumin alone is notoriously unreliable due to cytokine-driven hypoalbuminemia).

Complete Meta-Analysis Protocol Document

Preoperative Inflammatory-Nutritional Biomarkers and Anastomotic Leakage After Colorectal Cancer Surgery: A Systematic Review, Meta-Analysis, and Comparative Diagnostic Accuracy of Albumin, CRP, and the CRP-to-Albumin Ratio

PART 1: EXPLICIT ELIGIBILITY CRITERIA - NATURE AND SCOPE OF THE REVIEW

1.1 Nature of the Review

• Prognostic arm: pooled OR/RR for AL risk when a biomarker is above vs. below a threshold
• Diagnostic accuracy arm: pooled sensitivity, specificity, AUC, LR+, LR- for each biomarker and for CAR vs. albumin vs. CRP individually

1.2 Population (P) - Full Explicit Criteria

1.3 Exposure/Index Test (I) - Three Parallel Biomarkers

• Measured preoperatively (within 30 days)
• Reported as continuous variable OR as binary (hypoalbuminemia vs. normal, any threshold)
• Standard threshold: <3.5 g/dL (35 g/L) most common; also extract <3.0 and <2.5 g/dL
• Assay: BCG or BCP (record and use as subgroup variable)

• Measured preoperatively (within 30 days of surgery)
• NOTE: distinguish from postoperative CRP - these are fundamentally different exposures
• Reported as continuous variable or binary (elevated vs. normal, study-defined)
• Standard threshold: >5 mg/L (upper limit of normal) or study-defined
• Assay: standard CRP vs. high-sensitivity (hsCRP) - must subgroup

• Calculated as: CRP (mg/L) ÷ albumin (g/L). Units must be standardised
• OR as: CRP (mg/dL) ÷ albumin (g/dL) - these give different numeric values, requiring standardisation before pooling
• If not explicitly reported, extract CRP and albumin from the same blood draw and compute it yourself (if raw data or means/SDs are reported)
• Reported continuous or binary (above/below study-defined cut-off)

1.4 Comparator (C)

• Normal biomarker level (below threshold) as the comparator for binary analyses
• For diagnostic accuracy: any biomarker level (continuous spectrum - no explicit comparator required in ROC analysis)
• For comparative analysis: albumin and CRP serve as comparators against CAR

1.5 Outcome (O) - Primary and Secondary

• Clinical signs (fever, peritonitis, purulent/fecal drainage)
• Radiological evidence (CT or contrast study)
• Operative findings at re-laparotomy or endoscopy
• ISREC classification (any grade A+B+C)

• Overall postoperative morbidity (Clavien-Dindo ≥II)
• Major morbidity (Clavien-Dindo ≥III)
• Surgical site infection (SSI)
• 30-day all-cause mortality
• ICU admission
• Hospital length of stay

1.6 Study Design (S)

• Prospective cohort studies
• Retrospective cohort studies
• Nested case-control studies (if extractable sensitivity/specificity data)
• RCTs in which preoperative biomarker data are reported alongside AL outcome

• Cross-sectional studies (cannot establish temporal precedence)
• Case reports, case series <50 patients
• Conference abstracts without full peer-reviewed data
• Review articles and letters
• Studies with no raw or extractable data on both biomarker level and AL outcome

PART 2: FEASIBILITY - AVAILABLE STUDIES AND DATA SOURCES

2.1 Confirmed Primary Literature

2.2 Estimated Study Pool

• Studies reporting preoperative albumin + AL in CRC: 30-50 eligible studies
• Studies reporting preoperative CRP + AL in CRC: 20-35 eligible studies
• Studies reporting preoperative CAR + AL in CRC: 8-15 eligible studies (the scarcest - verify before PROSPERO registration)
• Studies reporting all three from the same cohort: 5-10 studies

2.3 Data Sources for Search

PART 3: HETEROGENEITY ANALYSIS

3.1 Why Heterogeneity Will Be High - and Why That Is Acceptable

3.2 Complete Taxonomy of Heterogeneity Sources

3.3 Statistical Heterogeneity Management

• CAR cut-off value (continuous moderator)
• Year of publication
• Proportion of rectal cancer patients
• Proportion receiving neoadjuvant therapy
• Country (Asia vs. Europe vs. Americas)
• Study quality score (NOS total)

PART 4: EXPLICITNESS, TRANSPARENCY, AND REPLICABILITY

4.1 Explicitness

4.2 Transparency

1. Full search strings in supplementary material: Every exact search string (MeSH terms + free text) for every database, exactly as run, with the date of search
2. PRISMA flow diagram: Every step from database hits through deduplication, title/abstract screening, full-text assessment, final inclusion - with reasons for exclusion at full-text stage
3. Pre-registered data extraction form: Must be declared in PROSPERO (template in Appendix)
4. Independent dual extraction: Two reviewers extract independently, third resolves disagreements - state this explicitly and state who the reviewers are
5. Risk of bias table: Individual QUIPS domain ratings for every included study, visually presented as a traffic light table
6. Code availability: If R/Stata code is used for analysis, deposit it on OSF (Open Science Framework) or GitHub and cite the repository in your Methods
7. GRADE summary table: Certainty of evidence per outcome per biomarker, formally graded

4.3 Replicability

PART 5: CRITICAL APPRAISAL - RISK OF BIAS ASSESSMENT

5.1 Tool Selection - Three Tools for Three Different Study Purposes

5.2 QUIPS - For the Prognostic Arm

5.3 QUADAS-2 - For the Diagnostic Accuracy Arm

5.4 Overall Risk of Bias Presentation

• Present a traffic light table (green/amber/red per domain per study) using the robvis R package
• Calculate proportion of studies at low/high/unclear risk per domain
• Conduct sensitivity analysis restricted to low-risk studies (those with multivariable adjustment AND pre-specified or externally validated cut-offs)
• Do NOT exclude high-risk studies from the primary analysis - include but flag

PART 6: MAIN OBSTACLES IN CONDUCTING THIS META-ANALYSIS

Obstacle 1: CAR Formula Inconsistency (MOST CRITICAL)

• CRP in mg/L ÷ Albumin in g/L → values typically 0.1-10
• CRP in mg/dL ÷ Albumin in g/dL → values typically 0.001-1
• CRP in mg/L ÷ Albumin in g/dL → incorrect mixing of units

Obstacle 2: Circular Cut-off Validation

Obstacle 3: Predominance of Postoperative Rather Than Preoperative CAR Data

Obstacle 4: Inconsistent AL Definitions

Obstacle 5: Confounding by Indication - The Fundamental Limitation

Obstacle 6: Lack of External Validation in Most Primary Studies

Obstacle 7: Small Primary Study Sizes

PART 7: CAN A MEDICAL STUDENT DO THIS?

Honest Assessment: Yes, With Significant Caveats

What a Medical Student CAN Do Independently

What Requires Supervisor/Biostatistician Input

Specific Challenges for a Medical Student

Practical Recommendations for a Medical Student

1. Secure a supervisor first - a clinician familiar with colorectal surgery AND a methodologist/biostatistician. A meta-analysis with neither is not publishable.
2. Start with a scoping search to confirm the primary literature volume before committing to PROSPERO registration
3. Use RevMan for screening and basic meta-analysis; R with metafor and mada for the diagnostic accuracy component
4. Two-reviewer system: Find a co-reviewer (another student or junior doctor) for all screening and extraction tasks. This is both methodologically required and practically important for quality control.
5. Register on PROSPERO early (before running the full search) - this protects the work and is required for publication
6. Budget: Allow 6-9 months from search to submission realistically
7. Target journals first: Decide target journal before writing, as word limits and format requirements vary significantly

PART 8: PROSPERO PROTOCOL TEMPLATE

═══════════════════════════════════════════════════════════════
PROSPERO REGISTRATION TEMPLATE
International Prospective Register of Systematic Reviews
═══════════════════════════════════════════════════════════════

REVIEW TITLE
Preoperative inflammatory-nutritional biomarkers and anastomotic 
leakage after colorectal cancer surgery: a systematic review, 
meta-analysis, and comparative diagnostic accuracy of albumin, 
C-reactive protein, and the CRP-to-albumin ratio

ANTICIPATED OR ACTUAL START DATE
[DD/MM/YYYY]

ANTICIPATED COMPLETION DATE
[DD/MM/YYYY] (allow 12 months from search date)

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
CONTACT DETAILS FOR THIS REVIEW

Review contact name: [SURNAME, First name]
Affiliation: [Institution, Department, City, Country]
Email: [professional institutional email]
ORCID: [XXXX-XXXX-XXXX-XXXX]

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
REVIEW TEAM

Name 1: [First author - student/junior researcher]
  Role: Principal investigator, search, screening, extraction, writing
  Affiliation: [Institution]

Name 2: [Supervisor - senior clinician]
  Role: Clinical expert input, protocol development, manuscript review
  Affiliation: [Institution]

Name 3: [Methodologist/biostatistician]
  Role: Statistical analysis, risk of bias, GRADE assessment
  Affiliation: [Institution]

Name 4: [Second reviewer for screening and extraction]
  Role: Independent dual reviewer
  Affiliation: [Institution]

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
BACKGROUND

Anastomotic leakage (AL) is among the most feared complications 
following colorectal cancer (CRC) resection surgery, occurring in 
approximately 3-11% of patients and accounting for up to 40% of 
postoperative mortality. Preoperative identification of patients 
at elevated risk of AL allows targeted interventions including 
nutritional optimisation, modification of surgical strategy, and 
prophylactic defunctioning stoma placement.

Serum albumin and C-reactive protein (CRP) are readily available, 
inexpensive preoperative laboratory markers that reflect nutritional 
reserve and systemic inflammatory status, respectively. The 
CRP-to-albumin ratio (CAR), calculated as CRP [mg/L] divided by 
albumin [g/L], integrates both dimensions and has been proposed as 
a superior composite biomarker. Prior single-centre studies suggest 
CAR outperforms either constituent alone in predicting postoperative 
complications, but no systematic review or meta-analysis has 
examined the comparative preoperative diagnostic accuracy of these 
three biomarkers specifically for AL after CRC surgery.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
REVIEW QUESTION

Primary question:
Is a high preoperative CAR, compared with normal CAR, associated 
with increased risk of anastomotic leakage in adults undergoing 
elective colorectal cancer resection with primary anastomosis?

Secondary questions:
1. Is preoperative hypoalbuminemia associated with increased risk 
   of AL after elective CRC resection?
2. Is elevated preoperative CRP associated with increased risk of 
   AL after elective CRC resection?
3. Which biomarker - albumin, CRP, or CAR - demonstrates superior 
   diagnostic accuracy (AUC, sensitivity, specificity) for predicting 
   AL in this population?
4. What is the optimal preoperative CAR cut-off for AL prediction 
   as determined by pooled ROC analysis?

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
SEARCHES

The following electronic databases will be searched:
- MEDLINE (via PubMed) from January 2000 to present
- Embase (via Ovid) from January 2000 to present
- Cochrane Central Register of Controlled Trials (CENTRAL)
- CINAHL (via EBSCO)
- Web of Science (Core Collection)
- Scopus
- ClinicalTrials.gov
- Reference lists of all included studies will be hand-searched

No language or publication status restrictions will be applied 
at the search stage. Non-English articles will be assessed using 
certified translation for Asian-language papers and 
machine-assisted translation (verified) for European languages.

Search date: [DD/MM/YYYY]

Sample search strategy (MEDLINE/PubMed):
("colorectal neoplasms"[MeSH] OR "colorectal cancer"[tiab] OR 
"colon cancer"[tiab] OR "rectal cancer"[tiab] OR "colorectal 
carcinoma"[tiab])
AND
("anastomotic leak"[MeSH] OR "anastomotic leakage"[tiab] OR 
"anastomotic leak"[tiab] OR "anastomotic fistula"[tiab] OR 
"anastomotic failure"[tiab] OR "anastomotic insufficiency"[tiab] 
OR "anastomotic complication"[tiab])
AND
("albumin"[MeSH] OR "serum albumin"[tiab] OR "hypoalbuminemia"[MeSH] 
OR "hypoalbuminaemia"[tiab] OR "hypoalbuminemia"[tiab] OR 
"C-reactive protein"[MeSH] OR "C reactive protein"[tiab] OR 
"CRP"[tiab] OR "CRP albumin ratio"[tiab] OR "C-reactive protein 
to albumin"[tiab] OR "inflammatory nutritional"[tiab] OR 
"preoperative biomarker"[tiab])

Full search strings for all databases will be reported in a 
supplementary file accompanying the final manuscript.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
ELIGIBILITY CRITERIA

Types of study
Eligible: prospective cohort, retrospective cohort, nested 
case-control, and randomised controlled trials reporting 
preoperative biomarker data alongside anastomotic leakage outcomes.

Ineligible: case series (<50 anastomoses), cross-sectional 
studies, conference abstracts, letters, editorials, narrative 
reviews, and animal studies.

Types of participants
- Adults ≥18 years
- Confirmed colorectal cancer (any histological subtype)
- Elective curative-intent colorectal resection with formation 
  of a primary colorectal anastomosis
- Emergency surgery excluded
- Metastatic-only populations excluded
- Non-CRC indications (IBD, diverticulitis) excluded unless CRC 
  patients are reported separately with extractable data

Types of index test (exposure)
Any of the following, measured preoperatively within 30 days 
before surgery:
(a) Serum albumin concentration (g/dL or g/L)
(b) Serum CRP concentration (mg/L or mg/dL)
(c) CRP-to-albumin ratio (CAR), calculated as CRP [mg/L] 
    divided by albumin [g/L] or equivalent

Types of reference standard / outcome
Primary outcome: Anastomotic leakage within 30 days of surgery, 
defined by clinical, radiological, or operative findings, 
or classified using ISREC criteria (any grade; Grade B/C in 
sensitivity analysis)

Secondary outcomes:
- 30-day all-cause mortality
- Overall postoperative morbidity (Clavien-Dindo ≥II)
- Major morbidity (Clavien-Dindo ≥III)
- Surgical site infection
- Unplanned re-operation
- Hospital length of stay

Minimum follow-up: ≥21 days

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
STUDY SELECTION

Stage 1 (Title/Abstract screening):
Two independent reviewers ([Name 1] and [Name 4]) will screen 
all titles and abstracts using Rayyan or Covidence software. 
A third reviewer ([Name 2]) will resolve disagreements. 
Interrater agreement will be reported using Cohen's κ statistic 
(target κ ≥0.70 before proceeding).

Stage 2 (Full-text assessment):
The same two reviewers will independently assess full texts of 
all potentially eligible studies against explicit eligibility 
criteria. Reasons for exclusion at this stage will be documented 
for each excluded study and reported in the PRISMA flow diagram.

Author contact:
Where studies appear eligible but report insufficient data, 
up to two contact attempts will be made to corresponding authors 
(2-week response window each).

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
DATA EXTRACTION

Two reviewers will independently extract data into a 
pre-specified Excel extraction form (template deposited on 
OSF prior to data collection). Disagreements will be resolved 
by the third reviewer.

Data extracted will include:
- Study identification: first author, year, country, journal
- Study design and period
- Sample size (total enrolled; number with anastomosis)
- Indication for surgery (CRC only vs. mixed)
- Tumour location (colon vs. rectum vs. mixed)
- TNM stage distribution
- Proportion receiving neoadjuvant therapy
- Surgical approach (open/laparoscopic/robotic)
- Diverting stoma rate
- Biomarker type, timing, and assay method
- Biomarker threshold used (pre-specified vs. ROC-derived)
- AL definition and ascertainment method
- AL rate (events/total patients with anastomosis)
- Unadjusted OR/RR/HR (95% CI) for AL
- Adjusted OR/RR/HR (95% CI) for AL + variables adjusted for
- Sensitivity and specificity (if ROC data reported)
- AUC and 95% CI (if reported)
- Youden index and optimal cut-off (if reported)

For studies reporting multiple biomarkers from the same cohort: 
extract all three (albumin, CRP, CAR) with corresponding 
diagnostic accuracy data for comparative analysis.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
RISK OF BIAS ASSESSMENT

Two independent reviewers will assess risk of bias for all 
included studies using the following tools:

1. QUIPS (Quality In Prognosis Studies, Hayden et al. 2013) 
   for studies included in the prognostic/risk factor analysis. 
   Domains assessed: (1) study participation; (2) study 
   attrition; (3) prognostic factor measurement; (4) outcome 
   measurement; (5) confounding measurement and accounting; 
   (6) statistical analysis and reporting.

2. QUADAS-2 for studies included in the diagnostic accuracy 
   analysis. Domains assessed: (1) patient selection; 
   (2) index test; (3) reference standard; (4) flow and timing; 
   plus applicability concerns for each of the first three domains.

Overall risk of bias will be rated as low, moderate, or high 
per domain, with narrative justification for each rating. 
Results will be presented as a traffic light summary table 
(generated using `robvis` in R).

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
STRATEGY FOR DATA SYNTHESIS

Quantitative synthesis (meta-analysis) will be performed when 
three or more studies report the same biomarker, outcome, and 
sufficient statistical data.

Prognostic arm:
- Effect measure: odds ratio (OR) with 95% confidence interval
- Model: random-effects (REML estimator)
- Software: R (`metafor` package)
- Heterogeneity: I², τ², and 95% prediction interval reported

Diagnostic accuracy arm:
- Bivariate random-effects model (Reitsma et al.)
- Outputs: pooled sensitivity, pooled specificity, SROC curve, 
  DOR, LR+, LR-
- Software: R (`mada` package) or Stata (`metandi`)
- Threshold effect: Spearman correlation of logit(sensitivity) 
  and logit(1-specificity) to test for threshold effect

Comparative accuracy:
- For studies reporting ≥2 biomarkers from same cohort, 
  direct comparison of AUC values using DeLong method 
  (within-study comparison)
- Pooled AUC from random-effects meta-analysis per biomarker

Publication bias:
- Funnel plot asymmetry assessed visually when ≥10 studies 
  per analysis
- Egger's test for prognostic arm
- Deeks' funnel plot test for diagnostic accuracy arm
- If asymmetry detected: trim-and-fill correction

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
SUBGROUP AND SENSITIVITY ANALYSES

The following subgroup analyses are pre-specified:
1. Surgery type: colon vs. rectum
2. Neoadjuvant therapy: yes vs. no
3. Cut-off source: pre-specified vs. internally ROC-derived
4. AL definition: ISREC-graded vs. clinical/other
5. Geographic region: Asian vs. Western cohorts
6. CRP assay: standard vs. high-sensitivity (hsCRP)
7. Study quality: NOS ≥6 vs. <6 (for QUIPS-rated studies)
8. Diverting stoma use: ≥20% vs. <20% of cohort

The following sensitivity analyses are pre-specified:
1. Restrict to studies with multivariable-adjusted OR only
2. Restrict to studies with preoperative biomarker measured 
   within 14 days of surgery (vs. 30-day window)
3. Restrict to studies with ISREC-defined AL only
4. Leave-one-out analysis to identify influential studies
5. Restrict to studies published 2010 or later (ERAS era)
6. Restrict to studies with ≥100 anastomoses

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
META-REGRESSION

Meta-regression will be performed when ≥10 studies are available 
per analysis, using the following covariates:
- CAR cut-off value (continuous)
- Proportion of rectal cancer patients (%)
- Proportion receiving neoadjuvant therapy (%)
- Year of publication (continuous)
- Geographic region (categorical)
- Mean study sample size
- Overall NOS/QUIPS score

Results will be presented as regression coefficients with 95% CI 
and interpreted with caution given multiple testing.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
EVIDENCE CERTAINTY - GRADE

The certainty of evidence will be formally assessed using the 
GRADE (Grading of Recommendations Assessment, Development 
and Evaluation) framework for each primary and secondary 
outcome per biomarker. For the diagnostic accuracy arm, the 
GRADE-approach for diagnostic test accuracy reviews will be 
applied. Evidence will be rated as High, Moderate, Low, or 
Very Low certainty. A Summary of Findings table will be 
prepared for each major outcome.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
REPORTING

This systematic review will be reported in accordance with:
- PRISMA 2020 (Page et al., BMJ 2021)
- PRISMA-DTA (McInnes et al., BMJ 2018) for the diagnostic 
  accuracy component
- MOOSE (Meta-analysis Of Observational Studies in 
  Epidemiology) for the observational study component

The PRISMA checklist will be included as a supplementary file 
with the final publication.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
CONFLICTS OF INTEREST

[State any conflicts or: "The authors declare no conflicts 
of interest relevant to this systematic review."]

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
FUNDING

[State funding source or: "This systematic review received 
no specific funding. All authors contributed in an 
unfunded voluntary capacity."]

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
STAGE OF REVIEW AT TIME OF REGISTRATION

[Select the most accurate option from PROSPERO dropdown:
- Review not yet started
- Ongoing (searches completed, screening started)
- Ongoing (screening completed, extraction started)
- Completed but not published]

Note: PROSPERO strongly recommends registration BEFORE 
executing the full electronic search. If searches have 
already begun, declare this honestly.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
RELATED REVIEWS

[State any related reviews you are aware of:]
- Elfadil U et al. 2026 (PMID 41913878): examined nutritional 
  status broadly (including albumin among multiple indices) and 
  AL in colorectal surgery; did not examine CRP or CAR; no 
  diagnostic accuracy analysis
- Heuvelings DJI et al. 2024 (PMID 39111814): examined 
  reporting quality of AL definitions; did not examine biomarkers
- Vun T et al. 2025 (J Clin Med): examined peritoneal fluid CRP 
  for postoperative AL prediction; different sample and timing 
  of biomarker

The present review is distinct from all above in examining 
preoperative biomarkers and including comparative diagnostic 
accuracy of albumin vs. CRP vs. CAR.

═══════════════════════════════════════════════════════════════
END OF PROSPERO TEMPLATE
═══════════════════════════════════════════════════════════════

Quick-Reference Summary Card

Easier, High-Impact Meta-Analysis Topics in GI Surgery / Oncosurgery

WHAT IS ALREADY SATURATED (Do Not Touch)

RECOMMENDED TOPICS - Ranked by Overall Feasibility Score

🥇 RANK 1 - TOP RECOMMENDATION

Preoperative Anaemia (Haemoglobin Level) and Anastomotic Leakage After Elective Colorectal Cancer Surgery: A Systematic Review and Meta-Analysis

• Collagen hydroxylation (oxygen-dependent process) - directly relevant to anastomotic healing
• Neutrophil oxidative burst - reduces anastomotic infection resistance
• Tissue oxygen delivery at the anastomotic site

🥈 RANK 2

Preoperative HbA1c Level and Risk of Anastomotic Leakage After Colorectal Cancer Surgery: A Systematic Review and Meta-Analysis

• Advanced glycation end-products that cross-link collagen → impairs anastomotic tensile strength
• Microvascular disease → reduced anastomotic perfusion
• Impaired neutrophil chemotaxis → higher anastomotic infection risk
• This is a causal story, not just confounding - much more defensible than albumin

🥉 RANK 3

Modified Glasgow Prognostic Score (mGPS) and Postoperative Complications After Colorectal Cancer Surgery: A Systematic Review and Meta-Analysis

RANK 4

Preoperative Blood Transfusion and Anastomotic Leakage After Colorectal Cancer Surgery: A Systematic Review and Meta-Analysis

RANK 5

Anastomotic Leak After Colorectal Cancer Surgery and Long-Term Oncological Outcomes: A Systematic Review, Meta-Analysis, and Update

• The exposure (AL yes vs. no) is extremely clean and binary
• The outcome (5-year overall survival, disease-free survival, local recurrence) is well-defined in every oncology study
• Hazard ratios are routinely reported (allowing pooling using standard log-HR methods)
• RevMan handles this entirely
• NOS for risk of bias
• No diagnostic accuracy component needed
• No threshold heterogeneity (AL yes/no)

MASTER COMPARISON TABLE

My Final Opinion

1. Zero competition - no published meta-analysis, no PROSPERO registration found
2. WHO-standardised exposure - Hb in g/dL, measured identically everywhere, no formula dispute, minimal heterogeneity in the exposure itself
3. Large primary literature - haemoglobin is reported in every CRC baseline table; you will find 50+ eligible studies without difficulty
4. Clinically actionable - directly informs preoperative iron optimisation protocols and surgical timing decisions
5. Technically simple - binary exposure, RevMan-executable, NOS risk of bias, no diagnostic accuracy arm needed
6. Strong biological mechanism - oxygen-dependent collagen synthesis makes this a causal story, not just confounding
7. ESPEN 2025 + ERAS guideline relevance - both recommend preoperative haemoglobin optimisation; your meta-analysis provides the quantitative foundation that these guidelines cite

ctDNA as a Prognostic Tool in GI Surgery: Topic-by-Topic Comparison

Preliminary verdict (before the tables)

PART 1 - The Saturated Zone (Do Not Enter)

PART 2 - The Open Niches (Genuine Gaps)

🥇 RANK 1 - BEST OPPORTUNITY

ctDNA as a Predictor of Pathological Complete Response (pCR) and Organ Preservation Eligibility in Rectal Cancer After Total Neoadjuvant Therapy (TNT): A Systematic Review and Meta-Analysis

🥈 RANK 2

ctDNA After Hepatocellular Carcinoma (HCC) Resection as a Predictor of Recurrence and Overall Survival: A Systematic Review and Meta-Analysis

🥉 RANK 3

Preoperative ctDNA Positivity and Risk of Early Recurrence After Curative-Intent Surgery for Colorectal Cancer: A Systematic Review, Meta-Analysis and Subgroup Analysis by Stage

RANK 4

ctDNA-Guided Adjuvant Chemotherapy Decision-Making in Stage II Colorectal Cancer: A Systematic Review and Meta-Analysis of Clinical Utility

RANK 5

ctDNA After Resection of Cholangiocarcinoma or Biliary Tract Cancer: Prognostic Value - A Systematic Review and Meta-Analysis

PART 3 - Master Comparison Table

PART 4 - Cross-Domain Verdict

ctDNA topics overall are harder, not easier, than the previous recommendations.

Where ctDNA is genuinely worth pursuing:

1. The pCR/watch-and-wait topic (Rank 1) is a high-impact, novel clinical question that no meta-analysis has answered yet - but it requires a supervisor with rectal cancer expertise and understanding of TNT protocols. Impact if published: likely Colorectal Disease, Diseases of Colon and Rectum, or Annals of Surgical Oncology.
2. The HCC hepatectomy topic (Rank 2) is an empty niche with growing primary literature and is technically simpler than the pCR topic. This is the best ctDNA topic for a student with a hepatobiliary surgeon as supervisor.
3. The preoperative-only CRC ctDNA topic (Rank 3) scores highest of all ctDNA options at 35/40 because it reframes existing data in a clinically novel way without requiring complex diagnostic accuracy methods.

Bottom line:

PROSPERO PROTOCOL + SEARCH GUIDE

Preoperative Anaemia and Risk of Anastomotic Leakage After Elective Colorectal Cancer Surgery: A Systematic Review and Meta-Analysis

SECTION A - PROSPERO REGISTRATION RECORD

FIELD 1: REVIEW TITLE

FIELD 2: REVIEW QUESTION

FIELD 3: BACKGROUND / RATIONALE

FIELD 4: PICOS ELIGIBILITY CRITERIA

• Benign colorectal disease (diverticulitis, inflammatory bowel disease, familial adenomatous polyposis without cancer)
• Emergency surgery (bowel obstruction, perforation, haemorrhage)
• Hartmann's procedure without anastomosis
• Paediatric populations (<18 years)
• Non-anastomotic surgery (abdominoperineal resection without anastomosis as the sole procedure)
• Studies where colorectal cancer patients cannot be disaggregated from non-cancer patients

• Overall postoperative complication rate (Clavien-Dindo grade II or above)
• 30-day mortality
• Length of hospital stay
• ICU admission rate
• Stoma formation (permanent or temporary, attributable to AL)
• Perioperative blood transfusion requirement

• Case reports and case series (n<20)
• Non-original research (review articles, editorials, conference abstracts without full data)
• Animal studies
• Studies reporting only bivariate/unadjusted associations without multivariate analysis, unless included for sensitivity analysis

FIELD 5: EXPOSURE MEASUREMENT

FIELD 6: SEARCHES

1. MEDLINE (via PubMed) - from inception to date of search
2. Embase (via Ovid) - from inception to date of search
3. Cochrane Central Register of Controlled Trials (CENTRAL) - from inception
4. Web of Science Core Collection - from inception
5. Scopus - from inception
6. ClinicalTrials.gov - for unpublished trial data
7. WHO International Clinical Trials Registry Platform (ICTRP)
8. PROSPERO - to identify relevant related protocols
9. Grey literature: Google Scholar (first 200 results), conference proceedings of the European Society of Coloproctology (ESCP) and American Society of Colon and Rectal Surgeons (ASCRS) 2019-2026
10. Reference lists of all included studies and relevant systematic reviews (backward citation tracking)
11. Forward citation search on Harada 2025 (PMID 39797981) and Bruns 2019 (PMID 31188183) using Google Scholar

FIELD 7: STUDY SELECTION

FIELD 8: DATA EXTRACTION

FIELD 9: RISK OF BIAS ASSESSMENT

FIELD 10: STATISTICAL ANALYSIS AND SYNTHESIS

• I² <25%: low heterogeneity
• I² 25-50%: moderate heterogeneity
• I² 50-75%: substantial heterogeneity
• I² >75%: considerable heterogeneity

1. Anaemia severity (mild vs. moderate vs. severe)
2. Cancer site (colon vs. rectal vs. combined)
3. Surgical approach (open vs. minimally invasive)
4. Defunctioning stoma use (yes vs. no)
5. Neoadjuvant therapy proportion (>50% vs. ≤50% of study population)
6. Study quality (NOS score ≥7 vs. <7)
7. Study design (prospective vs. retrospective)
8. Geographic region (Asian vs. European vs. North American)

1. By anaemia severity: mild (Hb 10-12.9 g/dL women; 10-12.9 g/dL men) vs. moderate (Hb 8-9.9 g/dL) vs. severe (<8 g/dL)
2. By cancer site: rectal cancer only vs. colon cancer only vs. combined
3. By surgical approach: open vs. laparoscopic/robotic
4. By defunctioning stoma use: protected vs. unprotected anastomosis
5. By neoadjuvant therapy: neoadjuvant chemoradiotherapy-exposed vs. naive
6. By risk of bias: low risk (NOS ≥7) vs. high risk (NOS <7)

• Excluding retrospective studies (prospective only)
• Excluding studies using non-WHO anaemia definitions
• Including only studies using adjusted ORs
• Excluding studies with NOS <6

FIELD 11: CERTAINTY OF EVIDENCE

FIELD 12: TYPE AND METHOD OF REVIEW

FIELD 13: LANGUAGE

FIELD 14: COUNTRY

FIELD 15: ANTICIPATED OR ACTUAL START DATE

FIELD 16: ANTICIPATED COMPLETION DATE

FIELD 17: FUNDING SOURCES / CONFLICTS OF INTEREST

FIELD 18: REVIEW TEAM

FIELD 19: JUSTIFICATION FOR DIFFERENCES FROM SIMILAR REGISTERED REVIEWS

SECTION B - FULL PROTOCOL DOCUMENT

B1. INTRODUCTION AND BACKGROUND

• World Health Organization. Haemoglobin concentrations for the diagnosis of anaemia and assessment of severity. WHO/NMH/NHD/MNM/11.1. Geneva: WHO, 2011.
• Bao et al. Int J Surg 2026 (PMID 40968737) - umbrella review showing anaemia NOT included as a pooled AL factor
• Harada et al. Surgery Today 2025 (PMID 39797981) - multivariate OR 9.94 for moderate/severe anaemia and AL
• Bruns et al. Dis Colon Rectum 2019 (PMID 31188183) - multicentre cohort, rectal cancer, anaemia associated with postoperative course
• Tsokkou et al. PMC 2025 (PMC12651639) - SR on preop anaemia optimisation in CRC, focused on iron supplementation NOT on AL as outcome
• Dias et al. Rev Col Bras Cir 2022 (PMID 36449942) - general preoperative risk factor MA, anaemia not specifically pooled
• ESPEN guidelines on clinical nutrition in surgery 2021 (Weimann et al. Clin Nutr 2021)
• ERAS Society guidelines for colorectal surgery (Gustafsson et al. World J Surg 2019)

B2. OBJECTIVES

1. To determine whether the risk of AL differs by anaemia severity (mild, moderate, severe)
2. To compare AL risk in colon cancer vs. rectal cancer surgery separately
3. To explore whether surgical approach (open vs. minimally invasive) modifies the anaemia-AL relationship
4. To examine the effect of defunctioning stoma use on the anaemia-AL association
5. To assess the certainty of evidence using GRADE

B3. PICOS CRITERIA (full detail)

B4. SEARCH STRATEGY

B5. STUDY SELECTION PROCESS

B6. DATA EXTRACTION FORM

B7. RISK OF BIAS - QUIPS DOMAINS (GUIDANCE)

SECTION C - COMPREHENSIVE SEARCH STRINGS

C1. MEDLINE (PubMed)

(
  (
    "anemia"[MeSH Terms] OR "anemia"[tiab] OR "anaemia"[tiab] OR
    "iron deficiency anemia"[MeSH Terms] OR "iron deficiency anaemia"[tiab] OR
    "haemoglobin"[tiab] OR "hemoglobin"[tiab] OR
    "low haemoglobin"[tiab] OR "low hemoglobin"[tiab] OR
    "haemoglobin level"[tiab] OR "hemoglobin level"[tiab] OR
    "preoperative haemoglobin"[tiab] OR "preoperative hemoglobin"[tiab] OR
    "preoperative anemia"[tiab] OR "preoperative anaemia"[tiab]
  )
  AND
  (
    "anastomotic leak"[MeSH Terms] OR "anastomotic leak"[tiab] OR
    "anastomotic leakage"[tiab] OR "anastomotic leakages"[tiab] OR
    "anastomotic dehiscence"[tiab] OR
    "anastomotic failure"[tiab] OR
    "colorectal anastomosis leak"[tiab] OR
    "rectal anastomosis leak"[tiab]
  )
  AND
  (
    "colorectal neoplasms"[MeSH Terms] OR
    "colorectal cancer"[tiab] OR "colorectal carcinoma"[tiab] OR
    "colon cancer"[tiab] OR "colon carcinoma"[tiab] OR
    "colon neoplasm"[tiab] OR "colonic cancer"[tiab] OR
    "rectal cancer"[tiab] OR "rectal carcinoma"[tiab] OR
    "rectal neoplasm"[tiab] OR "rectum cancer"[tiab] OR
    "sigmoid cancer"[tiab] OR "sigmoid carcinoma"[tiab] OR
    "colorectal surgery"[MeSH Terms] OR "colectomy"[MeSH Terms] OR
    "proctectomy"[tiab] OR "anterior resection"[tiab] OR
    "low anterior resection"[tiab] OR "ultra-low anterior resection"[tiab] OR
    "hemicolectomy"[tiab] OR "sigmoidectomy"[tiab] OR
    "right hemicolectomy"[tiab] OR "left hemicolectomy"[tiab] OR
    "total mesorectal excision"[tiab] OR "TME"[tiab] OR
    "colorectal resection"[tiab] OR "colonic resection"[tiab] OR
    "rectal resection"[tiab]
  )
)
NOT (
  "animals"[MeSH Terms] NOT "humans"[MeSH Terms]
)

C2. EMBASE (via Ovid)

1. anemia/
2. (anemia or anaemia or haemoglobin or hemoglobin or "low haemoglobin" or "low hemoglobin" or "preoperative anaemia" or "preoperative anemia" or "iron deficiency").ti,ab.
3. 1 OR 2
4. anastomotic leak/
5. ("anastomotic leak" or "anastomotic leakage" or "anastomotic dehiscence" or "anastomotic failure").ti,ab.
6. 4 OR 5
7. colorectal tumor/ or colon cancer/ or rectal cancer/ or colorectal surgery/
8. ("colorectal cancer" or "colon cancer" or "rectal cancer" or "colorectal neoplasm" or "colectomy" or "proctectomy" or "anterior resection" or "TME" or "total mesorectal excision" or "colorectal resection" or "hemicolectomy").ti,ab.
9. 7 OR 8
10. 3 AND 6 AND 9
11. limit 10 to NOT (animal not human)

C3. COCHRANE CENTRAL REGISTER OF CONTROLLED TRIALS (CENTRAL)

(anemia OR anaemia OR haemoglobin OR hemoglobin OR "preoperative haemoglobin" OR "preoperative hemoglobin")
AND
("anastomotic leak" OR "anastomotic leakage" OR "anastomotic dehiscence")
AND
("colorectal cancer" OR "colon cancer" OR "rectal cancer" OR "colorectal surgery" OR "anterior resection" OR "colectomy" OR "proctectomy")

C4. WEB OF SCIENCE CORE COLLECTION

TS=(
  (anemia OR anaemia OR haemoglobin OR hemoglobin OR "iron deficiency anemia" OR "preoperative anaemia" OR "preoperative anemia" OR "low hemoglobin" OR "low haemoglobin")
  AND
  ("anastomotic leak" OR "anastomotic leakage" OR "anastomotic dehiscence" OR "anastomotic failure")
  AND
  ("colorectal cancer" OR "colon cancer" OR "rectal cancer" OR "colorectal surgery" OR "anterior resection" OR "low anterior resection" OR "colectomy" OR "proctectomy" OR "hemicolectomy" OR "total mesorectal excision" OR TME)
)

C5. SCOPUS

TITLE-ABS-KEY(
  (anemia OR anaemia OR haemoglobin OR hemoglobin OR "preoperative anemia" OR "preoperative anaemia")
  AND
  ("anastomotic leak" OR "anastomotic leakage" OR "anastomotic dehiscence")
  AND
  ("colorectal cancer" OR "colon cancer" OR "rectal cancer" OR "colorectal surgery" OR "anterior resection" OR colectomy OR proctectomy)
)

C6. GREY LITERATURE AND SUPPLEMENTARY SEARCHES

• European Society of Coloproctology (ESCP) Annual Meetings 2019-2026
• American Society of Colon and Rectal Surgeons (ASCRS) Annual Meetings 2019-2026
• European Conference on Clinical Oncology (ECCO) 2022-2026

• Bao et al. Int J Surg 2026 (PMID 40968737)
• Dias et al. Rev Col Bras Cir 2022 (PMID 36449942)
• Heuvelings et al. Dis Colon Rectum 2024 (PMID 39111814) - AL reporting quality SR

• Harada et al. Surgery Today 2025 (PMID 39797981) - anchor primary study
• Bruns et al. Dis Colon Rectum 2019 (PMID 31188183) - anchor multicentre cohort

C7. SEARCH MANAGEMENT AND DEDUPLICATION

1. Export all search results to Zotero or Endnote in RIS/BibTeX format from each database
2. Import all records into Rayyan (free, cloud-based) or Covidence for duplicate removal and screening
3. Remove duplicates using the automated deduplication tool in Rayyan/Covidence followed by manual check
4. Document: total records retrieved per database, total after deduplication
5. Record exact search dates for each database (required for PRISMA 2020 flow diagram)

SECTION D - SUPPLEMENTARY PROTOCOL NOTES

D1. Handling Common Methodological Problems

D2. Anticipated Timeline

D3. Target Journals and Submission Strategy

D4. PROSPERO Submission Checklist

• All team members have ORCID iDs registered
• All team members have confirmed participation via PROSPERO email
• Conflict of interest field completed for each team member
• Full protocol PDF uploaded as attachment
• Search strategy in at least one database documented (PubMed string above)
• Guarantor has approved the record content
• Registration number will be assigned and published immediately on submission (no delay since Feb 2025 new platform)
• Record the CRD registration number and cite it in the Methods section of every subsequent manuscript

D5. PRISMA-P 2015 Compliance Checklist

1. Check PROSPERO one final time for competing registrations using: (anaemia OR anemia OR haemoglobin) NEAR colorectal NEAR anastomotic - the new PROSPERO search engine supports proximity operators
2. Run the PubMed search string (Section C1) in full before submission so you can report the approximate number of potential studies in Field 3 and confirm the gap is real
3. Add your actual names, ORCID iDs, and dates to Fields 15, 16, and 18 - PROSPERO will not publish without all team members confirming

PART 1 - HOW TO SEARCH PROSPERO PROPERLY

Step-by-Step PROSPERO Search Guide

Understanding How PROSPERO Search Works

The 5 Searches You Must Run (Run All of These)

(anemia OR anaemia OR haemoglobin OR hemoglobin) AND "anastomotic leak" AND colorectal

(anem* OR haemoglobin OR hemoglobin) AND (anastomot*) AND (colorectal OR colon OR rectal OR rectum)

(anemia OR anaemia OR haemoglobin OR hemoglobin):TI AND (anastomotic):TI

"anastomotic leak" AND "risk factor*" AND (colorectal OR "colorectal cancer" OR "rectal cancer")

"preoperative anemia" AND "colorectal cancer" AND surgery

How to Interpret Results

1. Read the review question field - is the outcome anastomotic leak specifically?
2. Read the PICOS criteria field - is the exposure specifically haemoglobin/anaemia, or is it something else (albumin, nutritional status, iron)?
3. Check the status - is the review "Ongoing", "Completed", or "Awaiting classification"? If "Completed" and published, search PubMed for the published paper.
4. Note the registration date - registrations from >3 years ago with no update are likely abandoned; note this and mention it in your PROSPERO record under Field 19 (justification for new review)

What You Are Looking For

Critical Finding from Pre-Search (Already Done)

CRD420251113455 (Tsokkou S, Konstantinidis I; registered July 2025; published MDPI Cancers 2025) Title: "Optimizing Preoperative Anemia in Non-Metastatic Colorectal Cancer: A Systematic Review on Surgical Recovery and Outcomes" Outcome: Iron supplementation strategies, IV vs. oral iron, transfusion requirements, haemoglobin optimisation, recovery outcomes NOT an outcome: Anastomotic leak is not the primary or secondary outcome of this review

PART 2 - IS THIS REMARKABLE? POTENTIAL AND OBSTACLES

A. Why This Could Be Genuinely Impactful

1. It is first-in-class at a pivotal moment

2. The Harada 2025 anchor study is a strong signal

3. It is directly actionable

4. Timing with the Patient Blood Management movement

5. The de Wit 2025 prospective study is perfect timing

B. Potential Obstacles in Any Oncology Setting

Obstacle 1 - Heterogeneity in the anaemia definition

• WHO criteria (Hb <13 g/dL men; <12 g/dL women) - the gold standard
• Hb <10 g/dL (moderate-severe only)
• Hb <11 g/dL or <11.5 g/dL (common clinical thresholds)
• Haematocrit <36% instead of Hb
• No explicit threshold at all - "anaemia" based on clinician judgement

Obstacle 2 - Anaemia type is rarely reported

Obstacle 3 - Confounding by indication is severe

• Have more advanced tumours (higher T/N stage, more bleeding)
• Be older and frailer
• Have more comorbidities (diabetes, renal failure)
• Receive neoadjuvant chemotherapy/radiotherapy more often (NCRT itself causes anaemia)
• Be nutritionally depleted (lower albumin, lower BMI)
• Have longer operating times (more complex surgery)

Obstacle 4 - Neoadjuvant therapy creates a subgroup problem

• Anaemia (bone marrow suppression from chemotherapy)
• AL risk (radiation-induced tissue fibrosis, impaired vascular supply)
• Altered haemoglobin trajectory (Hb may fall post-NCRT and then rise before surgery)

Obstacle 5 - Selection bias in defunctioning stoma use

C. Reasons Why This Might Get Rejected

Rejection Risk 1 - "Not specific enough"

Rejection Risk 2 - "This is just a sub-analysis of existing reviews"

1. Dias 2022 did not pool haemoglobin/anaemia as a specific exposure - they mentioned it as one of many factors without providing a pooled estimate
2. Your review is designed specifically to quantify the anaemia-AL relationship with subgroup analysis by anaemia severity, which no existing review does
3. You include more recent primary studies (Harada 2025, de Wit 2025, etc.) not available in 2022

Rejection Risk 3 - "Insufficient studies for meta-analysis"

• Include crude ORs in a sensitivity analysis if adjusted ORs are unavailable
• Contact authors of large cohort studies for subgroup data
• Perform a narrative synthesis alongside the meta-analysis if fewer than 5 studies with adjusted ORs are available

Rejection Risk 4 - "High heterogeneity with no explanation"

Rejection Risk 5 - "Low certainty of evidence undermines clinical applicability"

Rejection Risk 6 - Low-quality journal problem

Summary: The Honest Overall Assessment